The Mountain Car Problem - Design of Experiments for Reinforcement Learning - page 107

In general, we find that the ranges and locations of parameters that allow for con-
vergence are consistent with intuition and are explainable. However, once individual
subregions are explored on a more detailed level, we find that some subregions can
be quite simple with a single parameter that dominates performance, whereas others
can be quite complex with multiple parameters that have an effect on performance.
Additionally, in these complex subregions, the reasons why each of these parameters
have the effects that are seen is not intuitively obvious and will likely require further
experimentation.

References

Embrechts, M. J., Hargis, B. J., & Linton, J. D. (2010). An augmented efficient backpropagation
training strategy for deep autoassociative neural networks. In Proceedings of the 2010 Interna-
tional Joint Conference on Neural Networks (IJCNN), Barcelona, Spain, 18-23 July (pp. 1-6).
doi: 10.1109/IJCNN.2010. 5596828
Gatti, C. J., Embrechts, M. J., & Linton, J. D. (2013). An empirical analysis of reinforcement
learning using design of experiments. In Proceedings of the 21 st European Symposium on Arti-
ficial Neural Networks, Computational Intelligence and Machine Learning (ESANN), Bruges,
Belgium, 24-26 April (pp. 221-226). Bruges, Belgium: ESANN.
LeCun, Y., Bottou, L., Orr, G., , & Müller, K. (1998). Efficient backprop. In Orr, G. & Müller, K.
(Eds.), Neural Networks: Tricks of the Trade , volume 1524 (pp. 5-50). Berlin: Springer.
Moore, A. W. (1990). Efficient memory-based learning for robot control . Unpublished PhD
dissertation, University of Cambridge, Cambridge, United Kingdom.
Patist, J. P. & Wiering, M. (2004). Learning to play draughts using temporal difference learning
with neural networks and databases. In Proceedings of the 13th Belgian-Dutch Conference
on Machine Learning, Brussels, Belgium, 8-9 January (pp. 87-94). doi: 10.1007/978-3-540-
88190-2_13
Tesauro, G. (1992). Practical issues in temporal difference learning. Machine Learning , 8(3-4),
257-277.
Wiering, M. A. (2010). Self-play and using an expert to learn to play backgammon with temporal
difference learning. Journal of Intelligent Learning Systems & Applications , 2(2), 57-68.
Wiering, M. A., Patist, J. P., & Mannen, H. (2007). Learning to play board games using
temporal difference methods (Technical Report UU-CS-2005-048, Institute of Informa-

tion and Computing Sciences, Utrecht University). Retrieved from http://www.ai.rug.nl/ ~

mwiering/group/articles/learning_games_TR.pdf.